Today's AI Pulse
OpenAI launches ChatGPT Images 2.0 with multilingual text and real‑time web research
OpenAI has introduced ChatGPT Images 2.0, a next‑generation image model that adds multilingual typography, real‑time web research, and agentic reasoning to its generation pipeline. The new “Thinking” mode lets paid users produce up to eight coherent images per prompt, including floor plans, maps, UI mock‑ups, and apply capabilities to uploaded content.
Also developing:
Baidu Unveils Proprietary ERNIE 5 Beating GPT-5 Performance on Charts, Document Understanding and More
Baidu announced ERNIE 5.0, a proprietary omni‑modal foundation model that the company claims matches or outperforms OpenAI’s GPT‑5‑High and Google’s Gemini 2.5 Pro on benchmarks for document understanding, chart reasoning, vision‑language tasks, and language coding. The model is delivered through Baidu’s ERNIE Bot and Qianfan API at a premium price of $0.85 per 1K input tokens and $3.40 per 1K output tokens, roughly comparable to GPT‑5.1 but lower than Claude Opus. At the same event Baidu rolled out upgraded digital‑human services, no‑code tools, AI agents (GenFlow 3.0, Famou, MeDo, Oreate) and announced a push into international markets. Two days earlier it also released an open‑source multimodal model, ERNIE‑4.5‑VL‑28B‑A3B‑Thinking, adding a low‑cost alternative for developers.
Anthropic Details How It Measures Claude’s Wokeness
Anthropic unveiled new safeguards to make its Claude chatbot politically even‑handed, embedding system prompts that forbid unsolicited political opinions and using reinforcement learning to reward neutrality. The company also released an open‑source tool that scores Claude’s responses, reporting 95% even‑handedness...
Feed Drop: Into the Machine with Tobias Rose-Stockwell
The episode explores how AI’s rapid advancement—evident in tools like Claude 4.5 writing most code—creates dangerous outcomes and is driven by market‑dominance incentives rather than purely subscription revenue. Tristan Harris argues that AI companies prioritize user engagement and data collection to...
Hackers Use Anthropic’s AI Model Claude Once Again
Anthropic disclosed that Chinese state‑backed hackers leveraged its Claude large‑language model to automate roughly 30 cyber‑attacks on corporations and governments in September, with 80‑90% of the operations driven by AI. The attackers used Claude to generate commands, craft phishing content...
Upwork Study Shows AI Agents Excel with Human Partners but Fail Independently
Upwork’s Human+Agent Productivity Index evaluated three leading AI agents—Gemini 2.5 Pro, GPT‑5 and Claude Sonnet 4—on more than 300 real freelance jobs under $500. Working alone, the agents achieved modest completion rates, with the best‑performing model finishing only 64% of data‑science...
AI Hacks Echo Age‑old Human Cyberattack Tactics
Anthropic just shared deep details of a large-scale AI cybersecurity attack. Two hacks they exposed: 1) the hackers convinced Claude of a new role. It’s not an AI system made in Silicon Valley. No, of course not. It is an employee...
Google Labs' Pomelli Generates Custom Ads in Minutes
OK, Google Labs is cranking out some amazing work. Pomelli lets you go from your website link to custom ads you can tweak or regenerate in less than 3 minutes. Great for ideation. (Google needs a Tesla business model where...
OpenAI’s New LLM Exposes the Secrets of How AI Really Works
OpenAI unveiled an experimental weight‑sparse transformer, a tiny large‑language model built with sparse connections to make its internal logic observable. Though its performance is comparable only to early models like GPT‑1 and far below GPT‑5, the architecture forces features into...
Game Data Unlocks Real-World Insights and Competitive Edge
Amazing how valuable the game data is for real world data. Only games have action on controller and reaction in the world in a physics accurate way. And if you have data from thousands of games and tens of millions...
Mozilla Announces an AI ‘Window’ for Firefox
Mozilla announced the development of an AI‑powered "AI Window" for Firefox, an opt‑in browsing mode that integrates an AI assistant and chatbot. The feature will sit alongside private and classic windows, letting users select the underlying AI model and providing...
Even Realities’ New Smart Glasses Ditch Cameras and Speakers
Even Realities unveiled its G2 Display Smart Glasses, a human‑centric wearable that forgoes outward‑facing cameras and external speakers, relying only on four microphones and a micro‑LED/gradient waveguide display that creates a layered 3D floating spatial view. The device bundles AI...
Google Gemini Live Gets Its 'Biggest Update Ever' With 5 New Upgrades – Here's How to Try Them
Google has rolled out the "biggest update ever" to Gemini Live, its real‑time conversational AI, on Android and iOS. The upgrade adds five new capabilities – adjustable speaking speed, automatic tone modulation, a storytelling mode with character accents, seamless multi‑language switching,...
How Do You Redefine Research Delivery For The AI Era? Start With Trust.
In this episode, Forrester’s CTO discusses how AI is reshaping research delivery, emphasizing that trust is the non‑negotiable foundation for any generative‑AI tool. He explains the development of Izola, a client‑facing AI that pulls exclusively from Forrester’s own IP and...
Google’s NotebookLM Adds ‘Deep Research’ Tool, Support for More File Types
Google is adding a new "Deep Research" tool to its NotebookLM AI note‑taking platform, allowing the system to act as an autonomous researcher that creates a research plan, browses the web, and delivers a source‑grounded report that can be inserted...
Google’s NotebookLM Will Now Do ‘Deep Research’
Google announced that its NotebookLM AI workspace will integrate Gemini’s Deep Research agent, giving all users access within a week. Users can choose a fast or deep research mode, with the deep option delivering in‑depth analysis, source citations, and the...
AI-Driven Keynote Workflow: Design-Heavy vs Quick Turnaround
My current keynote preparation workflow with AI: IF DESIGN MATTERS Ex: a two-hour presentation to a client on AI trends in their industry - I pace and think about structure and flow and usually draft a few bullets in a google doc....
Algorithms Now Outpace Labs as Science’s Primary Tool
The most powerful scientific instrument of the 21st century isn't the electron microscope or the particle collider. It's the algorithm. Today, a scientist in biology, physics, chemistry etc. is more likely to be debugging a Python script than to be running...
Google’s SIMA 2 Agent Uses Gemini to Reason and Act in Virtual Worlds
Google DeepMind unveiled SIMA 2, an embodied AI agent that leverages the Gemini 2.5 flash‑lite model to reason, understand language, and act in virtual environments. Compared with its predecessor, SIMA 2 doubled task‑completion rates, demonstrating complex, commonsense behavior in games like No Man’s Sky and...
AI‑only University in Abu Dhabi Fuels Global World Models
World Models are shipping from around the world. From the only university that teaches only AI. I was there yesterday (it is in Abu Dhabi) and this school is so impressive. Have a class recorded by a professor there that I will...
AI Hype Scares Youth, but Human Roles Remain Essential
I recently received an email titled “An 18-year-old’s dilemma: Too late to contribute to AI?” Its author, who gave me permission to share this, is preparing for college. He is worried that by the time he graduates, AI will be...
LinkedIn Is Making It Easier to Search for People with AI
LinkedIn has introduced an AI‑powered people search that lets users describe the type of professional they need—e.g., “Northwestern alumni who work in entertainment marketing”—instead of relying on exact names, titles or companies. The system uses generative AI to surface and...
Inside LinkedIn’s Generative AI Cookbook: How It Scaled People Search to 1.3 Billion Users
LinkedIn has rolled out an AI‑powered people search that lets users pose natural‑language queries—e.g., “Who is knowledgeable about curing cancer?”—to a graph of 1.3 billion members. The system combines a 7‑billion‑parameter policy model distilled into a 1.7 B teacher and a 220 M...
LinkedIn Adds AI-Powered Search to Help Users Find People
LinkedIn is rolling out an AI‑powered people search feature that lets users enter natural‑language queries—such as “Find me investors in the healthcare sector with FDA experience”—to locate professionals more easily. The tool, initially available to premium U.S. members, expands on...
The Future of Disney Plus Could Involve AI-Generated Videos
Disney CEO Bob Iger said the company is "really excited about AI" and envisions Disney+ letting subscribers both create and consume short‑form AI‑generated videos. He also hinted at adding game‑like features through Disney’s partnership with Epic Games and turning the...
CDAO Responsibilities Are Evolving: Why AI Strategy Now Starts at the Top
The Chief Data and Analytics Officer (CDAO) is rapidly becoming a core C‑suite role as companies shift from data collection to AI‑driven operations. A recent survey shows 73.7% of firms now have a formal CDO/CDAO, up from 12% a decade...
AI Supercycle Depends on Upgraded Networks and Infrastructure
Why the AI Supercycle Will Fail Without Advanced Networks The next wave of AI isn’t just about models—it’s about the networks that support them. If connectivity, infrastructure and data flow aren’t upgraded, the “supercycle” may fall short. Read more 👉...
AI Masks True Talent, Turning Hiring Into Stagecraft
Hiring used to break because companies moved too slowly. Now it breaks because candidates move too fast. AI gave every applicant the ability to look qualified before they have proved anything. Resumes are AI-polished. Writing samples are AI-written. Coding tests...
Google DeepMind Is Using Gemini to Train Agents Inside Goat Simulator 3
Google DeepMind unveiled SIMA 2, a new video‑game‑playing agent that leverages the Gemini large‑language model to interpret text, voice or drawing commands and act on pixel‑level inputs across a range of 3D worlds, including Goat Simulator 3. Trained on human gameplay from...
SIMA 2 Shows Generalist AI Reasoning Across Games
Damn. DeepMind's generalist AI agent SIMA 2 evolved from basic instruction-following to actual reasoning companion. Uses vision and keyboard/mouse like a human player, works across dozens of games without touching game code. The robotics angle is obvious - if you...
Bridging AI Adoption Gap: From Startup Speed to Human Impact
The AI Innovation Gap AI adoption is now outpacing mobile phone uptake in the 2000s. Yet a troubling two-tier economy is emerging. In this clip from my conversation with Sasha Rubel (GenAI Policy Lead, AWS EMEA) and Alexandru Voica (Head...
Google Is Rolling Out Conversational Shopping—And Ads—In AI Mode Search
Google is rolling out conversational shopping in AI Mode search, letting U.S. users ask complex product queries and receive AI‑generated suggestions, guides, and tables powered by its Shopping Graph. The experience will include sponsored shopping content, while the Gemini app’s shopping...
Open‑source ESM Replacement Outperforms the Original
new open source drop in replacement for esm models that blows it out of the water, courtesy of @profluentbio
Switch to TOON, Cut LLM Tokens and Costs
If you’re still sending raw JSON into your LLMs, you’re burning tokens, latency, and budget! Try TOON (Token-Oriented Object Notation). Clear like YAML, compact like CSV: • 30–60% fewer tokens • Up to 50% lower costs • Shines for tabular data. Free and Open source 🧵↓...
Google Will Let Users Call Stores, Browse Products, and Check Out Using AI
Google is launching a suite of AI‑powered shopping tools for U.S. users ahead of the holiday season, adding conversational product search to Search’s AI Mode, an agentic "Let Google Call" feature that phones local stores on a shopper’s behalf, and...
Turn Kids' Sketches Into AI‑Generated Storybooks
Underrated AI Use Case for Parents: Take your kids sketches or drawings — ask the kids to give the characters names — upload them to ChatGPT and tell the AI to create a captivating children’s story based on them. Wait for...
GPT‑5.1 API Launch with 24‑Hour Prompt Caching
GPT-5.1 is now available in the API. Pricing is the same as GPT-5. We are also releasing gpt-5.1-codex and gpt-5.1-codex-mini in the API, specialized for long-running coding tasks. Prompt caching now lasts up to 24 hours! Updated evals in our blog post.
Google Augments AI Shopping with Conversational Search, Agentic Checkout and an AI that Calls Stores for You
Google announced a suite of AI‑powered shopping tools ahead of the holiday season, including conversational search in AI Mode, enhanced Gemini app responses, agentic checkout that can auto‑purchase items via Google Pay, and a Duplex‑based feature that calls local stores...
Sparse Circuits Reveal How Neural Networks Operate
Understanding neural networks through sparse circuits:
Separate Judgment From Routine to Reclaim Time
AI reveals the difference between work that needs judgment and work that just needs motion. The teams that separate the two gain time. The ones that mix them drown in noise.
Spotify’s New Audiobook Recap Feature Uses AI to Remind You of the Story so Far
Spotify is rolling out an AI‑driven "Recaps" feature in its iOS app that automatically generates a spoken summary of the portion of an audiobook a listener has already heard, after 15‑20 minutes of playback. The tool, initially limited to a...
Quick Poll: Choose Your Favorite GPT 5.1 Feature
Alright, GPT 5.1's been live for about a day now, so time for a quick poll! :) Which one gets your vote?
AI Masks Interview Gaps, Reveals Outdated Hiring Metrics
AI helps candidates answer questions they do not fully understand. For a while this makes everyone look stronger. Then it forces companies to confront how little their old signals ever measured in the first place.
Creating Marble’s Launch Video with Marble Technology
Check out how we used Marble to make the launch video for Marble! It’s so fun to work behind the scenes to use this technology to help create and tell stories!
Proofread ChatGPT‑written Articles Before Publishing.
If you're using ChatGPT to write your articles, at the very least read the output once before publishing it 😅 https://t.co/b7qYKA9unI
Creating Superintelligence Will Undermine, Not Amplify Power
If a country or company builds superintelligence, they'll and up *losing* power, not gaining power – this thought-provoking study explains why:
DeepMind's SIMA 2 Shows Incremental Yet Bittersweet Progress
Google DeepMind's SIMA 1 vs SIMA 2 The bitter lesson continues to be bitter sweet https://t.co/puwR92vCla
Coca-Cola's AI Christmas Ad Signals Creative Industry Shift
Coca-Cola will once again have its Christmas commercial generated entirely by AI this year. When a billion-dollar company like Coca-Cola is doing that, just think about the pressure currently weighing on the creative industry. #CocaColaXFes2025 #cocacola #AI #AIart #aivideo @CyberRobooo @CurieuxExplorer...
Deploy Custom Groq Models in FactoryAI in Under a Minute
how to set up custom models via @GroqInc in @FactoryAI droid in <1 min https://t.co/fK5NyjwZEu
Twitter Bots Start Posing Fabulous Questions in Comments
twitter bots are now asking fabulous questions in comments https://t.co/8PxtwWbrcv
Cursor's Funding Soars to $2.3B, Valuation $29.3B
cursor raises again!!!!! in june they raised $900 million at a $9.9 billion. Now they have raised $2.3 billion at 29.3 billion usd valuation 🤯 https://t.co/9jiDdJ76jW